Recognition of stacked base pairs

[blade2669]

Dear all,

I have a structure of parallel duplex anchored by i-motif (pdb is attached - C3GAGA_full.pdb). The step G4-A5 is particularly interesting for me because of cross-strand stacking of the bases (G4A5.pdb). When I run analyze command on whole structure (C3GAGA_full.pdb) I get total overlap area in the G4-A5 step equal to zero. I found some help in the topic "Base Stacking analysis" here on the forum (http://forum.x3dna.org/howtos/base-stacking-analysis/). I believe that following comment from xiangjun regarding the distance threshhold for recognition of stacked bases is critical to my problem:

 "The distance cutoff is 4.5 Å. If the shortest distance of any two-pair between the two sets of atoms is larger than the cutoff, then the stacking area is set to 0."

I also tried to run analyze on the extracted G4-A5 step (G4A5.pdb) which to my surprise gave me total overlap area of 12.04.

1) Is there any explanation why the overlap area in the G4-A5 step in the full structure is zero but in extracted structure is 12.04?
2) Is it possible that zero overlap is result of low threshold?
3) If so, is it possible to change the threshold and how?

Hopefully I attached all relevant files.

Thank you in advance, any help is much appreciated.

[xiangjun]

Hi,

Thanks for bringing up an insightful case of inconsistent characterization of base-stacking interactions in 3DNA v2.x. The details you provided are reproducible and helped me to detect where the problem is.

1) Is there any explanation why the overlap area in the G4-A5 step in the full structure is zero but in extracted structure is 12.04?
2) Is it possible that zero overlap is result of low threshold?
3) If so, is it possible to change the threshold and how?

The inconsistency is due to the different arrangement bases along the two chains. For C3GAGA_full_analyze.out, it is as below:

            Strand I                    Strand II          Helix
   1   (0.073) ....>A:...7_:[.DA]A-**+-A[.DA]:..21_:C<.... (0.065)     |
   2   (0.087) ....>A:...6_:[.DG]G-**+-G[.DG]:..20_:C<.... (0.071)     |
   3   (0.060) ....>A:...5_:[.DA]A-**+-A[.DA]:..19_:C<.... (0.069)     |
   4   (0.061) ....>A:...4_:[.DG]G-**+-G[.DG]:..18_:C<.... (0.072)     |
   5   (0.060) ....>A:...3_:[HCY]C-**+-C[.DC]:..17_:C<.... (0.049)     |

For G4A5_analyze.out, the bases are swapped, as shown below:

            Strand I                    Strand II          Helix
   1   (0.069) ...3>C:..19_:[.DA]A-**+-A[.DA]:...5_:A<...3 (0.060)     |
   2   (0.061) ...3>A:...4_:[.DG]G-**+-G[.DG]:..18_:C<...3 (0.072)     |

It turns out that this rearrangement is consequential in your case. If you're interested in knowing the technical details, please check the source code; particularly, the two functions with_stacking() and with_ss_overlap() in file ana_fncs.c.

You may give DSSR a try. With the --non-pair option, you'd get a consistent result for the overlap area of any two stacked bases. For example, run the command x3dna-dssr -i=C3GAGA_full.pdb --non-pair, you'd see the following:

Code: [Select]

List of 27 non-pairing interactions
.....
   6 A.DG4   C.DA19  stacking: 6.3(4.7)--pm(>>,forward) interBase-angle=1 min_baseDist=3.46
   7 A.DA5   A.DG6   stacking: 6.5(3.8)--pm(>>,forward) interBase-angle=15 connected min_baseDist=3.08
   8 A.DA5   C.DG18  stacking: 6.1(4.7)--mp(<<,backward) interBase-angle=7 H-bonds[1]: "O4'-N1(imino)[3.20]" min_baseDist=3.23
   9 A.DG6   A.DA7   stacking: 2.4(0.0)--pm(>>,forward) interBase-angle=7 connected min_baseDist=3.55
......
With the --analyze option, you can also get base-pair parameters as from 3DNA v2.x analyze program.

HTH,

Xiang-Jun

[xiangjun]

As a follow-up, I've updated 3DNA to v2.4.4-2019sep09. It fixes the issue of inconsistent stacking area you experienced. Using C3GAGA_full.pdb as an example, the output for the overlap area section is as below:

     step      i1-i2        i1-j2        j1-i2        j1-j2        sum
   1 AG/GA  1.87( 0.00)  0.00( 0.00)  0.00( 0.00)  6.49( 4.35)  8.36( 4.35)
   2 GA/AG  7.03( 4.09)  0.00( 0.00)  0.00( 0.00)  6.52( 4.42) 13.55( 8.51)
   3 AG/GA  0.00( 0.00)  6.09( 4.69)  5.95( 4.04)  0.00( 0.00) 12.04( 8.73)
   4 Gc/CG  7.10( 4.01)  0.00( 0.00)  0.00( 0.00)  6.11( 3.06) 13.21( 7.07)
   5 cC/cC  0.00( 0.00)  0.00( 0.00)  4.08( 1.06)  0.00( 0.00)  4.08( 1.06)
   6 CC/cc  0.00( 0.00)  3.84( 0.77)  1.04( 0.00)  0.00( 0.00)  4.88( 0.77)
   7 Cc/Cc  0.00( 0.00)  0.00( 0.00)  2.25( 0.64)  0.33( 0.00)  2.59( 0.64)
   8 cc/CC  0.00( 0.00)  2.66( 0.10)  1.49( 0.00)  0.04( 0.00)  4.19( 0.10)
   9 cC/cC  0.08( 0.00)  0.19( 0.00)  0.63( 0.00)  0.83( 0.00)  1.72( 0.00)
  10 CG/Gc  8.08( 4.05)  0.00( 0.00)  0.00( 0.00)  4.96( 2.33) 13.04( 6.38)
  11 GA/AG  0.00( 0.00)  5.76( 3.86)  6.23( 4.66)  0.00( 0.00) 12.00( 8.52)
  12 AG/GA  5.89( 3.84)  0.00( 0.00)  0.00( 0.00)  4.70( 2.58) 10.59( 6.42)
  13 GA/AG  5.59( 2.76)  0.00( 0.00)  0.00( 0.00)  3.24( 0.73)  8.83( 3.49)

Best regards,

Xiang-Jun

[blade2669]

I downloaded the new version of x3dna (v2.4.4-2019sep09) and I confirm that overlap area is now calculated correctly.

Thank you very much for showing me the root of the problem, comprehensive explanation and release of new version of x3dna.


With kind regards,
Aleš